Big Walk used AI to fix a bug. Now the indie gaming world is having a reckoning.

Big Walk developer House House said this week that LLMs were used sparingly late in development to help with technical problems and bugs. It put the AI-written share at well under 0.1% of the game's code. The disclosure came as 19 members of the 50-plus-member Independent Games Festival jury signed a statement declining to vote for any game using generative AI in any capacity.

The bug involved a Unity engine problem that produced "infinite length" shadows and slowed the game. A Unity engineer told the studio a fix wasn't possible, so it asked an LLM chatbot. The studio said it modified the suggestion extensively into 10 lines of code. At least one social media user says the bug was fixable with significant human effort.

The row shows how little AI use it takes to become contentious in indie games. Some judges said even minimal use outweighs their admiration for the game. Others argued that punishing a small, disclosed use discourages transparency and encourages hiding AI use. Valve's Steam rules do not require disclosure of code or chatbot use, while the IGF Awards make no such distinction.

The jury statement, released Monday, cited a significant increase in submitted games developed with generative AI tools. It named ecological, economic, psychological and labor harms and called the tools' use "not a foregone conclusion in our industry." It did not name any game, but Big Walk quickly became the focus after House House gave a statement to Aftermath, later added to the game's FAQ page.

House House wrote that its disclosure was not an endorsement of AI tools. It said it worried voluntary disclosures might contribute to normalizing the technology and apologized that its LLM usage may not meet the standards of those limiting generative AI's harm to games. Judges Austin Walker and Damon Reece praised the game but said minimal AI use still overrode that respect, with Reece calling it a pain that has to happen to steer the culture away from AI.

Others pushed back. Ryerson said many acted as if there were no levels to AI use, and Tarason said the developers spent years on the work and asked a chatbot for two minutes how to fix a bug. Professor Brendan Keogh argued that zero tolerance is counterproductive because it encourages people to hide use, while narrative designer Hannah Nicklin called zero tolerance a viable political position aimed at a system rather than individuals.